Natural frequencies improve Bayesian reasoning in simple and complex inference tasks

Area of Science:

• Cognitive Psychology
• Decision Science
• Statistical Reasoning

Background:

• Natural frequencies improve Bayesian inference compared to probabilities.
• Previous research focused on simple, dichotomous cue/hypothesis scenarios.
• Real-world problems often involve multiple cue values, hypotheses, or cues.

Purpose of the Study:

• To investigate if natural frequencies enhance Bayesian inference in more complex situations.
• To determine if learning natural frequencies for simple tasks transfers to complex tasks.

Main Methods:

• Study 1: Medical students performed Bayesian inference tasks with varying complexity (cue values, hypotheses, number of cues) using either natural frequencies or probabilities.
• Study 2: Participants were taught natural frequencies for simple tasks and then tested on complex tasks.

Main Results:

• Natural frequencies increased Bayesian inferences by an average of 37 percentage points across four complex conditions in Study 1.
• In Study 2, learning natural frequencies transferred to complex tasks, yielding 40% and 81% correct inferences for tasks with three cue values and two cues, respectively.

Conclusions:

• Natural frequencies significantly improve Bayesian reasoning beyond simple dichotomous situations.
• Training with natural frequencies facilitates transfer of learning to complex decision-making tasks.
• Natural frequencies are a more broadly applicable tool for enhancing statistical inference.